Background/Purpose:   Localized
scleroderma has both inflammatory and fibrotic components contributing to its
effect on the skin and underlying tissue.   The extent and duration of
inflammation during the active phase of the disease is thought to be the major
contributor to long-term disease damage and disability in LS.  Therefore, accurately
identifying the active disease state is imperative, which may not always be
apparent in subtypes with deeper tissue involvement or slowly progressive
changes, such as with the Parry Romberg subtype.  Current commercially
available serological markers of disease activity, such as ESR and CRP, are
rarely elevated in LS.  Therefore, identifying new circulating biologic markers
reflecting active disease status would be helpful for patient management;
especially when deciding when to initiate or discontinue therapy in unclear
cases. 

Methods:   Forty-five LS subjects
with ≥ 2 serial serum samples who had active disease (Physician Global
Assessment – Activity > 0) at the first sera collection and had subsequently
entered inactive disease state at follow-up (remission) were included in a 31
cytokine/chemokine Luminex panel (T_H1 / 2 / 17 associated panels)
run on 72 total LS subjects.   The cytokine profile of active vs. inactive
disease state was analyzed utilizing both statistical modeling and data mining
techniques.  Unadjusted and adjusted mixed effects logit models were fitted
using activity status as the outcome and each of the cytokines as the primary
predictor. Covariates for adjustment were included one at a time. Cytokine
levels were categorized into quartiles and Association Rules Mining (web-based
analyses) and Bayesian Network assessment were used as alternative analytical
strategies. 

Results:   In unadjusted analyses,
the cytokines/chemokines that were significantly elevated at the baseline
active disease visit compared to the inactive disease state at follow-up were
IP-10 (Odds Ratio (OR) [95% confidence interval] = 2.1 [1.4, 3.2], p
<0.001), TNF-α (OR = 1.8 [1.1, 3.0], p = 0.016), and MCP-1 (OR =
2.0 [1.1, 3.9], p = 0.034).  After adjusting one at a time for several
clinical covariates, such as age at first sample, disease duration, gender, mLoSSI,
LoSDI, ANA, Anti-ssDNA and AHA antibodies, IP-10 remained significant for all
adjustments, TNF-α significant for all except mLoSSI, and MCP-1
significant for all except disease duration.  Data mining approaches
demonstrated similar results with elevation of IP-10 and TNF-α in the
peripheral blood (4^th quartile) showing the strongest links in the
active disease node, followed by other T_H1 associated cyto/chemokines
and T_H17 associated cytokines.

Conclusion:   IP-10 and TNF-α
were identified as inflammatory mediators that predict disease activity when
comparing paired active and inactive LS serologic cytokine/chemokine profiles. 
These data support earlier findings of correlations of IP-10 and TNF-α
with disease activity parameters in a cross-sectional Luminex plasma study. 
These may serve to augment clinical decision making when disease activity
status is not straightforward by clinical examination.  Further study with
additional longitudinal serological data points is underway.